Imputation of missing data in photovoltaic panel monitoring system

Abstract

In scientific research, data acquisition and processing play a fundamental role. In photovoltaic systems, given their nature, this process presents deficiencies due to various factors such as the dispersion of the installed modules, climatic conditions or the amount of information that must be obtained, so the processes of data acquisition, storage and processing are very important. The present research developed a data acquisition, storage and processing system for photovoltaic systems, following the European standards IEC 60904 and IEC 61724 for data acquisition, Fog Computing for information storage and finally Machine Learning was used for processing. The results showed that the KNN-based model obtained a SCORE of 99.08%, MAE of 25.3 and MSE of 93.16. Concluding that the KNN-based model is the most robust model for data imputation in PV system monitoring.